def _set_align(rsc_src: Tuple[Aligner, dict, Dict[str, int]],
entries: List[Entry]):
"""
음절과 형태소 분석 결과를 정렬한다.
Args:
rsc_src: (Aligner, restore dic, vocab out) resource triple
Word: Word 타입
entries: 엔트리 리스트
"""
bad_entries = []
for entry in entries:
if entry.is_sharp:
continue
entry.left_align = align_patch(rsc_src, entry.raw,
Morph.to_str(entry.left))
if not entry.left_align:
entry.err_msg = 'fail to align left'
bad_entries.append(entry)
continue
entry.right_align = align_patch(rsc_src, entry.raw,
Morph.to_str(entry.right))
if not entry.right_align:
entry.err_msg = 'fail to align right'
bad_entries.append(entry)
continue
assert len(entry.left_align) == len(entry.right_align)
print_errors(bad_entries)
def run(args: Namespace):
"""
actual function which is doing some task
Args:
args: program arguments
"""
aligner = Aligner(args.rsc_src)
restore_dic = load_restore_dic('{}/restore.dic'.format(args.rsc_src))
if not restore_dic:
sys.exit(1)
vocab_out = load_vocab_out(args.rsc_src)
khaiii_api = KhaiiiApi(args.lib_path, args.rsc_dir, '{"errpatch": false}')
for line_num, line in enumerate(sys.stdin, start=1):
line = line.rstrip('\r\n')
if not line or line[0] == '#':
continue
raw, left, right = line.split('\t')
left_align = align_patch((aligner, restore_dic, vocab_out), raw, left)
if not left_align:
logging.info('invalid %d-th line: left align: %s', line_num, line)
continue
right_align = align_patch((aligner, restore_dic, vocab_out), raw,
right)
if not right_align:
logging.info('invalid %d-th line: right align: %s', line_num, line)
continue
if len(left_align) != len(right_align):
logging.info('invalid %d-th line: left/right diff: %s', line_num,
line)
continue
pos_cnt = 0
neg_cnt = 0
for sent in _sent_iter(args):
pos_cnt_sent, neg_cnt_sent = _cnt_pos_neg(
khaiii_api, raw, (left_align, right_align),
(aligner, restore_dic, vocab_out), sent)
pos_cnt += pos_cnt_sent
neg_cnt += neg_cnt_sent
if neg_cnt > 0:
break
if neg_cnt > 0 or pos_cnt == 0:
logging.info('invalid %d-th line: +%d, -%d: %s', line_num, pos_cnt,
neg_cnt, line)
continue
print('{}\t{}\t{}'.format(raw, left, right))